The BBC is reporting that the UK government is planning to pay publishers in return for making papers publicly available. According to David Willetts this will "herald a new era of academic discovery". The BBC story is based on a recent Report of the Working Group on Expanding Access to Published Research Findings which was accepted on the 16th of July. Whilst this looks like a step in the right direction, it is a missed opportunity to reform an out of date and broken system.

The current situation is that researchers do the research, write the papers, referee them, and then in most cases sign away the copyright to a publisher. The publisher then charges researchers to read the papers. The vast majority of the work involved in publishing papers is done by researchers at no cost to the publisher, effectively paid for by universities through staff time, and the publisher then charges universities a large amount every year to access the papers. Sounds crazy right? There are growing protests, but the hold the large publishing houses have over scientific research remains.

There are basically three functions of a journal: dissemination of papers, refereeing to check for originality an errors (which mainly benefits the reader), and reputation (which mainly benefits the author). The first of these, dissemination, was important in the days when papers had to be printed and bound, but is now irrelevant. The internet long ago made dissemination of papers instantaneous and essentially free; The last time I looked up a paper in a physical book was about 5 years ago, and that just because I was looking for something from the '50s for my thesis introduction.

The second purpose, refereeing, is done by other researchers, usually two other "experts" in the field. Like everyone else I occasionally referee papers so I know publishers don't pay anything for this, but it's just expected. Many of the arguments for the value journals add are based on refereeing. The problem is that refereeing is very limited, depends greatly on who the referees happen to be, and can't catch any but the most glaring errors or omissions. Checking a paper's method and results for errors (accidental or deliberate) happens through the scientific process of others trying to repeat them, not through a couple of people reading over the paper on a spare weekend. Just because a paper has been refereed is no guarantee that it's correct, as shown by the number of retractions and occasional cases of fraud. For readers, refereeing isn't working and probably isn't worth the trouble in my opinion. As an author, building or maintaining your reputation is a far greater incentive to make sure your papers are original and free of error. In short, if you publish bad papers others will rapidly stop reading them, and you will get no citations; If you commit fraud to get your results then eventually it will be discovered. Refereeing has very little to do with it.

The final purpose of journals, reputation, is really the only one which allows the journals to keep going. The kudos, career prospects and sometimes bonuses (though not at York I should point out) which come from publishing in high impact journals like Nature, Science, and PRL are a powerful incentive for researchers. Until a highly cited publication in PPCF is given the same weighting as a similarly cited publication in Nature, this is not going to change. This has been pointed out elsewhere, including in a good article It's Not Academic by Mike Taylor.

The papers I read (and so cite) most are those which appear on the arXiv.org e-print server started by Paul Ginsparg and run by Cornell University. It's free to publish, free to access, and there's no refereeing. They also run a charmingly anacronistic but very useful email alert system so every morning I get a list of abstracts in the areas I'm interested in. No weeks or months of waiting for a journal to publish; I can be reading a paper within 24 hours of the author uploading it. You might expect that the arXiv would be full of rubbish but it's not, or at least no more than most refereed journals. Like all papers you just have to read them with a bit of healthy scepticism. Unfortunately at the moment plasma physics isn't very well represented on arXiv; Today there were 45 submissions in High Energy Physics theory, 111 in Condensed Matter, 95 in Astrophysics, and a shameful 3 in plasma physics. Most journals allow preprints of papers (the version before refereeing) to be put on arXiv so I put mine on there, and try to encourage others to do so as well.

There are some signs of change, and this proposal will at least allow the people who have paid for the research (the tax-paying public) to see the results, but it won't help to fix the more basic problems. Researchers will continue to do most of the work refereeing and editing for the journals on a volunteer basis, then pay the same journals to put the result on their website. Given that the journals need papers to survive, and the vast majority of these come from publicly funded research, the Government could have used this strong position to negotiate a much better deal for taxpayers. Instead they seem to be going down the line of tweaking the current system whilst maintaining the publishers' profits.